ABSTRACT
O ver 3 decades of research on human infants has shown us that they enter the world with an impressive set of sensory abilities (e.g., Eimas, Siqueland, Jusczyk, & Vigorrito, 1971; Jusczyk, 1997; Kellman & Arterberry, 2006). Further, the most recent decade has made clear that infants are also equipped with a set of cognitive abilities that allows them to parlay their initial sensory experiences into the beginnings of adult systems, such as language and spatial reasoning (e.g., Gerken, 2005; Oakes, Ross-Sheehy, & Luck, 2006). Through a series of snapshots of infant abilities at different ages, infant researchers are beginning to map out the trajectory of development in different cognitive domains. They are also beginning to hypothesize what sort of learning mechanism(s) might be needed to connect the known points of the developmental path. Of particular interest are mechanisms that might be involved in learners’ ability to generalize from a limited set of experiences. Generalization allows learners to behave appropriately when they encounter new input that shares relevant properties with prior experiences, and, importantly, to discriminate new input that shares relevant properties with prior experiences from otherwise similar input that does not share those relevant properties. For example, generalizing from instances of the concept dog implies that the learner both behaves appropriately toward the new instances of dog (e.g., refers to them with the word dog) and behaves differently toward similar nondogs (e.g., does not refer to horses with the word dog).
